|Publication number||US6931580 B1|
|Application number||US 09/524,254|
|Publication date||Aug 16, 2005|
|Filing date||Mar 13, 2000|
|Priority date||Mar 13, 2000|
|Publication number||09524254, 524254, US 6931580 B1, US 6931580B1, US-B1-6931580, US6931580 B1, US6931580B1|
|Inventors||Kevin J. Barcomb, Leendert M. Huisman, Mark F. Olive, Kevin C. Quandt|
|Original Assignee||International Business Machines Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (16), Non-Patent Citations (1), Referenced by (4), Classifications (13), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates to a method for fail analysis of objects in integrated circuits and semiconductor wafers, and, more particularly, to a method for rapid fail analysis which links the fail data to layout information available about the device under test such that the physical location of a failing object is determined and may be visually plotted.
The manufacture of integrated circuits (ICs) on semiconductor wafers is a complex process and involves many different steps. Failure to correctly perform one or more of the steps on one or more wafers in a batch will likely cause the circuits on a wafer to fail. Sometimes, random circuit failures occur. Generally, there is little that can be done to avoid random failures. Moreover, random failures do not provide much information on the manufacturing process that can be used to enhance manufacturing yield.
However, failures are often distributed in a pattern as a result of one or more process steps being improperly carried out. These non-random failure patterns can serve as a guide to provide valuable information as to the sources of the failures. Thus, one of the goals of fail analysis is to get inventories of what objects on an IC actually failed along with the display, presentation, and analysis of those inventories. This information makes it possible to make a rapid identification of common failures, chip and wafer scale patterns of failures, as well as, if possible, an identification of the actual defect that caused the failures.
Fail analysis is done after manufacturing tests are performed on the IC and test data are obtained. Traditionally, fail analysis has been done only at two levels. At a first level of fail analysis, sort codes are assigned to the integrated circuits. The sort codes tell if the integrated circuit passed all the tests, or, if not, during which phase of the test the IC failed. The integrated circuits are then sorted, based on their quality, usually into two grade levels: good and reject. At the first test level, very little detail is available about what actually failed on the IC, and very little usable information is extracted from the fail data, other than the phase of the test at which the failure occurred. Consequently a second level of testing and analysis is sometimes necessary. At the second test level, detailed information is obtained, such as the generation of detailed bit maps of memory chips. These bit maps show which bits contained incorrect values at any phase of the memory test. The second level provides very detailed information, however, at the cost of an extended test session, in addition to the normal manufacturing test session, to extract the bit map data. Moreover, such a detailed analysis seems to be applicable only to memory devices.
Large volumes of wafers are manufactured every day and it is impractical to perform a detailed bit analysis of each failed object on a wafer. Therefore, in practice, only a few failed IC's are bit mapped. It is currently a manual process to choose which failing IC's to bit map by manually going through the fail data. Consequently, it is difficult to pick the failed IC's that are representative of the largest yield detractors.
U.S. Pat. No. 5,838,951 to Song describes a common method of forming a wafer map. After testing the dies or chips on the wafer, the dies are sorted into various grade levels and the positions of the dies on the wafer are recorded. The various grades of the dies on the wafer are then displayed. However, the '951 patent does not identify which objects on the dies failed or which particular device forming an object may have caused the fault.
Consequently, it is difficult to see wafer level problem areas when it is desired to select a particular failing object, such as an embedded array, to bit map. Thus, there is a need for a visual tool which gives a wafer level view of all failing objects.
A method for analyzing test data for objects on an IC or a wafer is provided. The test data is linked to available layout information about the object under test. Certain objects are selected based on the test data. A representation of the selected objects is placed on a map of the IC or on a map of the wafer. The representation should correspond to the physical location of the object on the IC or wafer. Preferably, the representation comprises one or more polygons that enclose all devices that make up the object.
The present invention will now be further describe in the following pages of specification when taken in conjunction with the attached drawings, in which:
The present invention provides a method for analyzing test data of an embedded object. For purposes of this invention, an embedded object is a set of electronic devices, like transistors and nets, that make up a structure with a recognizable function. Examples are embedded memory arrays, scan chains, macros, cores, etc. The object is embedded in the sense that it is surrounded by other circuitry on the chip, such that it is not directly accessible from the input and output terminals or pads of the chip. As a result, testing an embedded memory array, as well as other embedded objects, is difficult.
The invention is preferably used in conjunction with an initial level of testing. As described above, at this initial or first level, sort codes are assigned to the IC's that describe if the IC passed all the tests, or, if not, during which phase of the test the IC failed. For many embedded objects, the test sequence for the IC explicitly tests whether or not that object is operational, and the resulting fail data contain the information needed to tell whether that object passed its associated test. Examples of first level tests are scan tests that determine whether each scan chain on the IC actually operates correctly as a scan chain and embedded memory array tests that determine whether embedded memory arrays are defect free. These and other types of first level tests, or manufacturing tests, are well known to one skilled in the art and will not be described in detail here.
The method of the present invention extracts the maximal amount of information from the manufacturing test data. In other words, it can be used in conjunction with know manufacturing tests already being used today. However, the invention uses far more information than just what phase of the test caused the object to fail and provides far more information in return. This is accomplished by linking the fail data obtained from the manufacturing testing to available layout information about the object under test. Of course, the same techniques may also be applied when additional test sessions have produced additional fail data.
The present invention preferably shows the failure of an embedded object by placing a representation of that object on a map of the failing IC or on a map of the wafer that contains the failing IC. Ideally all failing objects are represented on the map. The representation should comprise one or more polygons that enclose all the devices that make up the embedded object. In the case of a memory array, the polygon may simply be a rectangle that encloses the bit cells, the sense amplifiers and pre-charge logic, and the address and decode logic. The polygons should be placed in their correct locations on the map of the IC or the wafer. That is, the polygons should be placed such that they would enclose the devices that make up the embedded object if those objects had, instead, been shown on the map of the IC or the wafer. This map may be an actual visual map or a set of coordinates from which a visual map could be produced.
Consequently, the present invention provides a much more detailed view of what actually failed on an IC or on a wafer, and makes much more extensive use of the existing fail data than the prior art. By using polygons rather than the actual shapes that make up the devices in the failed embedded object, it is ensured that the maps can be produced rapidly. Additionally, this allows a visual map to be plotted rapidly using existing wafer map software, such as Plyview™, if required.
Turning for a moment to
Initially, according to an embodiment of the invention, it should be determined which of the embedded objects on the semiconductor wafer are to be tested, as show in step 10 of
Turning now to step 18, the integrated circuits are tested. The test may be any of a number of known tests, for example, on LSSD scan test known to one skilled in the art. The test data output by the test is then analyzed to determine which object failed on which integrated circuit, per steps 20–22. These steps may include analyzing the test data to determine that a portion of an embedded object or device making up an embedded object failed and, from that determination, determining which of the embedded objects failed.
After the test is performed, the interpreted test data, including which embedded objects failed the test, is combined with the location of each embedded object. This enables the physical location of the failing objects on the IC or wafer to be established, per step 24. Preferably, this is accomplished by referring to the cross-reference map. On that map, the locations of the failed embedded objects on the wafer or IC are linked to their respective representative polygons. From this information the physical location of the failed objects can be determined. Next, according to step 26, an IC map or wafer map of the failing objects is generated. The map should physically plot the location and shape of each of the failed objects. This is done by placing a representation, i.e., the polygon, of the failed object on the map. The polygons are preferably arranged on the wafer map in the exact locations of the embedded objects they represent. That is, the polygons are placed such that they would enclose the devices making up their respective embedded object if those embedded objects had been, instead, shown on the wafer map. The wafer map can be an actual visual map, or a set of coordinates from which a visual map may be produced. As mentioned above, a visual map may be plotted rapidly using existing wafer map software, such as Plyview™, if required.
The map of
Next, the embedded objects, which caused the test results to be incorrect, are identified. This can be done referring to the correspondence between the objects and the failing latches, steps 34–36. In know scan tests, the test data applied to the IC is selected with reference to the function of the circuit. Thus, when a part of the test data is incorrect, a corresponding embedded object which failed may be identified. Preferably a one to one correspondence between an object and one or more failing latches in the test data is provided.
Per steps 38–40, the layout data of the IC is referred to calculate the locations and shapes of the failing objects. A cross-reference map as described above may be used for this purpose. The locations and shapes of the failing objects are then plotted on a wafer map. The wafer map, or several wafer maps may then be displayed and analyzed to detect fault patterns and possible causes of the faults, per steps 42–44. Once identified, the difficulties with particular process steps or in forming arrays on parts of the wafer can be identified and remedied, improving yield.
The present invention provides a much more detailed view of what actually failed on an IC or on a wafer and makes much more extensive use of the fail data obtained during the manufacturing test. The use of polygons rather than the shapes that make up the devices in the failed embedded object ensures that the fail maps can be produced rapidly and, if a visual map is required, be plotted rapidly using existing wafer map software such as Plyview™. Additionally, the present invention fills a gap when, for example, embedded memories fail but no bit map can be generated economically.
The foregoing description of the invention illustrates and describes the present invention. Additionally, the disclosure shows and describes only the preferred embodiments of the invention, but as aforementioned, it is to be understood that the invention is capable of use in various other combinations, modifications, and environments and is capable of changes or modifications within the scope of the inventive concept as expressed herein, commensurate with the above teachings, and/or the skill or knowledge of the relevant art. The embodiments described herein above are further intended to explain best modes known of practicing the invention and to enable others skilled in the art to utilize the invention in such, or other, embodiments and with the various modifications required by the particular applications or uses of the invention. Accordingly, the description is not intended to limit the invention to the form disclosed herein. Also, it is intended that the appended claims be construed to include alternative embodiments.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3961252||Dec 20, 1974||Jun 1, 1976||International Business Machines Corporation||Testing embedded arrays|
|US4460999||Jul 15, 1981||Jul 17, 1984||Pacific Western Systems, Inc.||Memory tester having memory repair analysis under pattern generator control|
|US4479214||Jun 16, 1982||Oct 23, 1984||International Business Machines Corporation||System for updating error map of fault tolerant memory|
|US5164666||Jul 10, 1985||Nov 17, 1992||ICT Integrated Circuit Testing Gesellschaft fur Halbeleiterpruftechnik mbH||Method and apparatus for analyzing errors in integrated circuits|
|US5240866||Feb 3, 1992||Aug 31, 1993||At&T Bell Laboratories||Method for characterizing failed circuits on semiconductor wafers|
|US5256578||Dec 23, 1991||Oct 26, 1993||Motorola, Inc.||Integral semiconductor wafer map recording|
|US5325309||Apr 30, 1991||Jun 28, 1994||Lsi Logic Corporation||Method and apparatus for integrated circuit diagnosis|
|US5430734||Feb 12, 1993||Jul 4, 1995||Metalithic Systems, Inc.||Fault-tolerant waferscale integrated circuit device and method|
|US5544308||Aug 2, 1994||Aug 6, 1996||Giordano Automation Corp.||Method for automating the development and execution of diagnostic reasoning software in products and processes|
|US5561293||Apr 20, 1995||Oct 1, 1996||Advanced Micro Devices, Inc.||Method of failure analysis with CAD layout navigation and FIB/SEM inspection|
|US5787190||Jun 27, 1997||Jul 28, 1998||Advanced Micro Devices, Inc.||Method and apparatus for pattern recognition of wafer test bins|
|US5828825||Apr 12, 1996||Oct 27, 1998||Intel Corporation||Method and apparatus for pseudo-direct access to embedded memories of a micro-controller integrated circuit via the IEEE test access port|
|US5831992||Oct 9, 1997||Nov 3, 1998||Northern Telecom Limited||Methods and apparatus for fault diagnosis in self-testable systems|
|US5838951||Aug 28, 1996||Nov 17, 1998||Anam Industrial Co., Ltd||Wafer map conversion method|
|US6128403 *||Feb 19, 1998||Oct 3, 2000||Mitsubishi Denki Kabushiki Kaisha||Wafer map analysis aid system, wafer map analyzing method and wafer processing method|
|US6785413 *||Aug 24, 1999||Aug 31, 2004||International Business Machines Corporation||Rapid defect analysis by placement of tester fail data|
|1||Lallier, K.W., et al., "Relating Logic Design To Physical Geometry In LSI Chip" , IBM(R) Technical Disclosure Bulletin, vol. 19, No. 6, Nov. 1976, pp. 2140-2143.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8234597||Jul 31, 2012||International Business Machines Corporation||Tool and method to graphically correlate process and test data with specific chips on a wafer|
|US8773944 *||Feb 8, 2012||Jul 8, 2014||Qualcomm Incorporated||Concurrent multiple-dimension word-addressable memory architecture|
|US20090183133 *||Jul 16, 2009||Flemming Mark J||Tool and method to graphically correlate process and test data with specific chips on a wafer|
|US20120134229 *||May 31, 2012||Chihtung Chen||Concurrent multiple-dimension word-addressable memory architecture|
|U.S. Classification||714/742, 382/149, 714/733|
|International Classification||G11C29/56, G11C29/00, G06F11/00|
|Cooperative Classification||G11C29/56, G11C29/56008, G11C29/006, G11C2029/0401|
|European Classification||G11C29/56B, G11C29/00W, G11C29/56|
|Mar 13, 2000||AS||Assignment|
Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:BARCOMB, KEVIN J.;OLLIVE, MARK F.;HUISMAN, LEENDERT M.;AND OTHERS;REEL/FRAME:010680/0211
Effective date: 20000310
|Feb 23, 2009||REMI||Maintenance fee reminder mailed|
|Aug 16, 2009||LAPS||Lapse for failure to pay maintenance fees|
|Oct 6, 2009||FP||Expired due to failure to pay maintenance fee|
Effective date: 20090816